Return To AVSS Home Page


Improving maternal and child health is an important public health objective in California and throughout the nation. Toward this goal, the Community and Organization Research Institute at the University of California, Santa Barbara, in cooperation with the Maternal and Child Health Branch of the California Department of Health Services, has developed and maintains the Maternal and Child Health (MCH) Data Base. The data base consists of numerous factors (52 of which are standard variables) related to maternal and newborn characteristics as well as pregnancy outcomes. Examples include maternal age and parity, prenatal care, previous pregnancy history, birth plurality, complications, cesarean section, birthweight, and perinatal mortality rates.

The data are derived from the linked vital records files compiled by the California Department of Health Services. Linking the death certificate with the corresponding birth certificate is an essential procedure, since data from the birth record may be used to supplant information that is not reported or inaccurately reported on the death certificate. Among other advantages, matching permits the standardization of perinatal mortality rates based on newborn risk characteristics. Standardized perinatal mortality rates can be used as one indicator of the effectiveness of perinatal care.

Using a five-year cohort of data from 1966 through 1970, a computerized data base was established at the University of California, Santa Barbara in the mid-1970s. Gradually, the data base was expanded to incorporate available linked record cohorts. The MCH Data Base presently contains the 1960 and 1965 through 1987 birth cohorts; more than eight million births. Beginning in 1978, the research methods have been subjected to professional peer review and reported in scientific journals., This work was the first known attempt to measure the effectiveness of hospital care on perinatal outcomes using vital records data.

The Maternal and Child Health Data Base Descriptive Narrative was first published in 1980, and with its appendix (the Statistical Appendix), has been updated each time a new cohort year becomes available from the state. This edition covers the 1986 through 1990 birth cohort.

We believe that the title Descriptive Narrative is appropriate since the purpose of this report is to organize and summarize the statistics on perinatal care for the births that occur in California. We hope that such basic information will facilitate further research by health care providers in their search for the causes and explanations of perinatal outcomes. Through careful observation of the patients, health care providers often gain insights and intuitions into specific problems. In this context the data reported herein should not be interpreted dogmatically, but are best viewed as guideposts to be used in conjunction with other sources of information. While population studies minimize the problems associated with small samples, there nevertheless remain some limitations. The MCH Data Base is complementary to, not a substitute for, more detailed information describing the perinatal care within a particular community. It is appropriate, therefore, for hospitals and medical groups to review medical records and other local information to better understand the statistics available in the MCH Data Base.

The original purpose of the MCH Data Base was to develop and test a method for measuring the effectiveness of medical care in hospitals. The impact of medical care on the patient population can be evaluated in terms of its efficacy, its effectiveness, or its efficiency. Efficacy measures whether medical procedures improve health outcomes under ideal conditions. But for a procedure to be effective, it must improve outcomes in real-world settings. Finally, a medical procedure is efficient insofar as it makes economic sense. Effectiveness is closely related to the concept of quality.

There are numerous aspects to quality. Just as there are different aspects of quality medical care, there are also different types of assessments. The taxonomy for quality measurement was first conceptualized by Avedis Donabedian in 1966. Following his guidelines, most observers agree that the quality of medical care can be measured on three distinct levels: structure, process, and outcome. Evaluation on the basis of the structure of medical care resources seeks to assess the adequacy of the environment in which medical care takes place. Measurement on the process level uses care itself as a basis to determine if medicine was properly practiced. Assessments based on outcomes have been termed the "ultimate" validations of the effectiveness and quality of medical care by Donabedian since structure and process influence outcome. Each type of measurement has advantages and disadvantages.

Structural Measures
Evaluation of health care on the basis of the structure of resources seeks to assess the adequacy of the environment in which medical care takes place. Physical facilities, equipment, the qualifications of medical staff, and organizational structure are taken into account. This type of assessment is used primarily for licensing and for hospital accreditation. The results generally are binary: either acceptable or unacceptable. Structural assessments have an advantage for the medical community: If a deficit is found, a solution will logically follow. For example, if a structural quality assessment concludes that a hospital is lacking a particular resource, then the solution of acquiring the necessary resource is readily apparent. However, a limitation of structural assessments is that it may not be sufficient to insure effective medical care.

Process Measures
Process measures use care itself as the indicator of quality by analyzing the activities of physicians or other heath care providers to determine if medicine is properly practiced. Assessments include individual chart reviews and observations of practice. For the medical community, this is probably the most commonly used type of quality assessment. Part of the reason this procedure is so commonly used is that process assessments can be educational. Chart reviews and mortality reviews can act as a forum for the exchange of ideas. Another advantage of process measures is that solutions often logically follow the detection of a problem. The detection of the problem, however, is limited to the expertise and training of the reviewers.

These assessments are limited by the inadequacies of medical records as to what procedures were actually performed, the differences in the judgement of medical professionals, and frequent inaccessibility of medical records by those outside of the medical facility being assessed. In addition, case investigations are often anecdotal and rarely are truly "blind" reviews. For mortality cases, process measures frequently become an exercise in determining if a particular death was preventable or unpreventable.

Outcome Measures
The third type of assessment is the outcome measure. The validity of outcome-based measures is seldom questioned since outcomes are direct measures of "health" itself. Outcome not only reflects the development of state-of-the-art medical technology, but also indicates the degree to which existing technology has been applied. Yet, like the other two types of assessments, outcome measures also have some disadvantages which can be frustrating to the medical community. Unlike structural and process measures, if a problem is detected with an outcome measure, the solution does not logically follow. Outcomes are dependent not only of medical care but also on genetic, environmental, and behavioral differences. There are often long delays before the impact of these factors are evident. Another limitation on outcome measures is that they are necessarily based on group results --- not individual cases, and thus do not indicate the quality of care delivered to an individual patient. When hospital staff are informed that an outcome study indicates that their hospital's mortality rate is unusually high, they frequently ask for information describing preventable deaths. But outcome studies usually do not provide that kind of detailed information.

Despite these limitations, the outcome measures of medical care have become increasingly employed. For example, Kaiser Permanente, HCFA, the American Hospital Association, American College of Obstetrics and Gynecology (ACOG), the Vermont-Oxford Trial Facilitation Service, the National Perinatal Information Center, and the Joint Commission for Accreditation of Healthcare Organizations have all developed, or are in the process of developing, outcome assessments of medical care. The use of "Risk Adjusted Outcome Measures" (RAMO) has increased in no small part from the work by Mark Blumberg, who has not only done original research in the field, but has also contributed useful review articles and coined the term RAMO. In February, 1991 the California Association of Hospitals and Health Systems (CAHHS) convened a conference on health care data, with special emphasis on severity adjusted outcome measures. Partly as a result of the CAHHS-sponsored conference, Assembly Bill 524 introduced by Assembly Member Bronzan was passed by the California Legislature and signed into law by Governor Wilson. It refers to RAMO and lists three major objectives:

1. To promote and conduct risk-adjusted outcome studies.
2. To strengthen the existing state data collection effort and data base, by deliberate incremental additions or changes to, and expansion of, the abstract record required by current law.
3. Create an ongoing technical advisory capability to provide guidance to the state on advancement and refinement of outcome measures and studies which are feasible to purchasers, and economical to produce.

These objectives were realized with the recent publication of the first annual report of the California Hospital Outcomes Project.

It should be noted that the MCH Data Base was developed as a by-product of research attempting to find an area of medical care that would be most amenable to outcome measures of care. Reviewing some of the reasons that perinatal mortality was chosen as a basis for outcome measurement will illustrate some of the relevant issues when considering the implementation of an outcome measure.


In a Strategy For Evaluating Health Services, David Kessner and his coworkers described the "tracer" methodology for evaluating health services as follows:

The tracer concept was borrowed from the formal sciences. In physiology, for example, scientists use radioactive tracers to study how a body organ such as a thyroid gland handles a critical substance such as iodine. ... In measuring organ functions, or processes, of a health care body the tracers are discrete, identifiable health problems that flow through the system, each shedding light on how particular parts work, not in isolation, but in the system.

Kessner and his colleagues list six criteria for the selection of health service tracers; they should:
1. Have a significant functional impact on the health of the individual.
2. Be easy to diagnose.
3. Have reasonably high prevalence rates.
4. Be sensitive to the quality or quantity of health care services received by the patient.
5. Have well-defined techniques of medical management.
6. Have well-understood social, cultural, economic, and behavioral factors.

These six criteria are largely satisfied by using mortality as a perinatal outcome, since:
1. The perinatal period is one of high risk of adverse outcome where the effects of medical care are likely to be observable.
2. By using mortality as the perinatal outcome, there is no question regarding its diagnosis.
3. Childbirth is a universal event that constitutes 10-20 percent of hospital admissions; however, the low probability of perinatal death (about 0.1 percent) does pose a potential small number problem.
4. The historic decline in perinatal mortality rates in association with improvements in medical technology is proof of the responsiveness of perinatal outcomes to medical care.
5. In spite of rapid technological change in medical science, there is a general consensus concerning the optimal treatment of the perinate. Modern communication channels and continuing education in the medical professions foster homogeneity of medical care protocols.
6. A review of the epidemiology of perinatal mortality reveals a unique situation regarding the measurement of the impact of confounding variables such as nutrition, genetics, and environment. Briefly, the multiple effects of a large number of confounding variables can be subsumed by a single universally-measured quantitative index: birthweight. Moreover, birthweight is highly predictive of birth outcomes, especially for perinatal mortality.

Thus, perinatal mortality is well-suited to satisfying the Kessner criteria as a health services tracer. The following are some additional advantages of using perinatal mortality as a RAMO measure:
1. Births and deaths are single events. A birth cannot occur at one hospital and then again at another; nor can a death occur twice. Thus, there is no problem posed by multiple admissions or outcomes.
2. Infant mortality has traditionally been a standard indicator of the adequacy of health care.
3. Births and deaths are reliably reported. State law requires that health care providers report the occurrence of all births and deaths.
4. Birth and death data are readily accessible to qualified researchers. Moreover, the content and reliability of vital statistics is regularly monitored, thereby ensuring more accurate and complete data.
5. Births and deaths have been matched since 1970 in California to create an annual linked birth-death cohort file. This makes it possible to study the relationship between perinatal risk factors and mortality outcome.

These are but some of the reasons why perinatal mortality as reported by birth and death certificates serves well as a RAMO measure of the effectiveness of medical care. The MCH Data Base uses the concept of a standardized mortality ratio to apply vital records data to risk adjusted outcome-based health services assessment. The statistical model used in the MCH Data Base recognizes that the greatest difficulty in using outcomes for quality assessment is in patient risk differentials. For example, over 80 percent of the variance in perinatal mortality rates between hospitals can be explained by newborn factors and less that 20 percent by differences in other factors. It overcomes this inherent difficulty in a RAMO approach that utilizes the unique role of birthweight as both a proxy for maternal-fetal health as well as a powerful predictor of perinatal outcome.

Updated September 18, 1996 by RL Williams

Return To AVSS Home Page